CN1155189C

CN1155189C - A Simplified Method of Two-layer Weighted Parallel Interference Cancellation Method

Info

Publication number: CN1155189C
Application number: CNB011355271A
Authority: CN
Inventors: κ��÷; 魏立梅
Original assignee: Huawei Technologies Co Ltd
Current assignee: Huawei Technologies Co Ltd
Priority date: 2001-10-09
Filing date: 2001-10-09
Publication date: 2004-06-23
Anticipated expiration: 2021-10-09
Also published as: CN1411189A; US7321581B2; EP1443694A4; EP1443694A1; DE60218329D1; EP1443694B1; ATE354894T1; US20040246927A1; WO2003032545A1; DE60218329T2

Abstract

The present invention puts forward a simplified algorithm of a two-layer weighted parallel interference cancellation algorithm, which keeps the performance of the two-layer weighted parallel interference cancellation algorithm and simultaneously greatly reduces operation volume. Because hyperbolic tangent calculation is related in the two-layer weighted parallel interference cancellation algorithm, the two-layer weighted parallel interference cancellation algorithm can hardly realize the hyperbolic tangent calculation. For solving the problem, the simplified algorithm of the two-layer weighted parallel interference cancellation algorithm of the present invention uses a piecewise linear judgment method or a table lookup method to replace hyperbolic tangent judgment; the substance is to use a piecewise linear judgment function L(x) or the judgment function T(x) of the table lookup method to approach a hyperbolic tangent function tanh(x).

Description

Method for simplifying double-layer weighted parallel interference cancellation method

Technical Field

The invention relates to a multi-user detection technology of a base station in a CDMA mobile communication system, in particular to a parallel interference cancellation method in the CDMA system.

Background

CDMA systems have been the development direction of third generation mobile communications due to their advantages of high capacity, high quality of service, and good security. Multiple Access Interference (Multiple Access Interference) limits the improvement in CDMA system capacity and performance. The single-user receiver cannot eliminate the influence of multiple access interference on user signal detection, and the detection performance of the receiver is reduced under the conditions of increased number of users and near-far effect. Designing a receiver resistant to multiple access interference is the key to taking advantage of the high capacity and high quality of service of the CDMA system.

The multi-user detection technique is an enhanced technique for overcoming the influence of multiple access interference and improving the capacity of a CDMA system. The method can make full use of the information of a plurality of users to carry out joint detection on the signals of the plurality of users, thereby reducing the influence of multiple access interference on the performance of a receiver as much as possible and improving the capacity of a system.

Verdu proposed in 1986 a multi-user detector based on maximum a posteriori probability, i.e. a maximum likelihood sequence detector. Although this detector is an optimal detector, it is highly complex and requires an estimate of the received signal amplitude and phase information. This makes the maximum likelihood sequence detector difficult to apply. Therefore, sub-optimal multi-user detection methods must be studied.

Sub-optimal multi-user detection methods are roughly divided into two categories: a linear detection method and an interference cancellation method. The linear detection method performs a linear transformation on the soft output of the single-user detector to produce a set of new outputs that improve performance. Such methods mainly comprise: a decorrelation Detector (Decorrelating Detector), a minimum mean Square Error Detector (minimum mean Square Error Detector), a Polynomial Expansion Detector (Polynomial Expansion Detector), and the like. The linear detection method has good performance, but the calculation is complex. The interference cancellation method treats the signal of the desired user as a useful signal and treats the signals of other users as interference signals; the interference of other users is eliminated from the received signal to obtain the signal of the expected user, and then the signal of the expected user is detected, thereby improving the performance of the system. The interference cancellation method can be divided into: serial Interference Cancellation (Serial Interference Cancellation) and Parallel Interference Cancellation (Parallel Interference Cancellation). The successive interference cancellation method sequences the user signals in descending power order. Firstly, the user with the maximum power is judged and detected, then the user signal is regenerated, the signal of the user is removed from the received signal, and the detection of other users is not interfered by the user signal. And then, detecting the user signal with the second largest power, and regenerating and eliminating the signal of the user with the second largest power, so that the detection of the rest users is not interfered by the user with the second largest power. And then removing the interference of other users from the received signal according to the sequence. The performance of the method is better than that of a single-user detector, but the method has the disadvantages of large delay, power sequencing, large calculation amount and sensitivity to initial signal estimation. The parallel interference cancellation method cancels the signal interference of all other users for each user in parallel from the received signal. The method has the advantages of better performance than a single-user detector, small time delay and small calculation complexity, and is the most possible method at present.

The parallel interference cancellation method has a large performance improvement relative to a single-user detector under a high signal-to-noise ratio, but has a reduced performance improvement relative to the single-user detector under a low signal-to-noise ratio. In a CDMA system, power control can compensate for the fading characteristics of the channel to some extent, allowing the system to operate at a lower signal-to-noise ratio to increase the capacity of the system as much as possible. Therefore, how to improve the performance of the parallel interference cancellation method under the condition of lower signal-to-noise ratio has important significance.

The performance of the parallel interference cancellation method can be effectively improved by the partial parallel interference cancellation method. Different from the traditional parallel interference cancellation method, the method comprises the following steps: the traditional parallel interference cancellation method completely eliminates the multiple access interference to which the user is expected to be subjected from the received signal; and the method for partly parallel interference cancellation sets a weight value for each stage of interference cancellation, weights the multiple access interference suffered by the expected user, and only partly eliminates the multiple access interference in the process of interference cancellation. Michael Buehrer and strong p. nicoloso published a news report on "Partial Parallel Interference Cancellation for CDMA" by the institute of electrical and electronic engineering, fifth generation 1999 (IEEE transactions on Communications, pp.658-661, vol.47, No.5, 1999). This paper was obtained from a theoretical analysis: under the Gaussian channel, the traditional parallel interference cancellation method completely eliminates the multiple access interference suffered by the expected user from the received signal, and the estimation of the signal of the expected user is biased estimation in the situation; the partial parallel interference cancellation method only partially cancels the multiple access interference, can correct the deviation of the signal estimation of the expected user, and leads the judgment result to be more reliable. Under the condition of lower signal-to-noise ratio, the performance of the partial parallel interference cancellation method is obviously superior to that of the traditional parallel interference cancellation method.

A weighted parallel interference cancellation method based on bayesian criterion is disclosed by US patent 5418814, which is also a weighting method. The method is different from the weighting basic principle of a partial parallel interference cancellation method, and is a bit-level weighting method based on the minimum mean value of decision cost. The method sets a cost function of judgment, determines a reliability coefficient of a judgment result of each bit by taking the minimum mean value of the judgment cost as a criterion, and performs bit-level weighting on a signal regenerated by the bit by using the coefficient, so that the interference generated by the bit of a user is only partially eliminated in the elimination of the multiple access interference. Compared with the traditional parallel interference cancellation method, the method has the advantages that the performance is improved, and particularly under the condition of low signal-to-noise ratio, the performance is obviously improved.

Although the two methods effectively improve the performance of the traditional parallel interference cancellation method under the condition of lower signal-to-noise ratio, the improvement amplitude is limited. The double-layer weighted parallel interference cancellation method combines a partial parallel interference cancellation method and a weighted parallel interference cancellation method based on a Bayesian rule, so that the method performance is further improved, and the method performance is greatly improved particularly under the condition of low signal-to-noise ratio.

A two-tier weighted parallel interference cancellation method is described below.

The structure of the double-layer weighted parallel interference cancellation receiver is shown in fig. 1, and the internal structures of the PIC structure 1 and the PIC structure 2 at the last stage are respectively shown in fig. 2 and fig. 3. The first-stage PIC structure takes the baseband signal of the received signal as the input signal of each user and processes the baseband signal to obtain the output signal of each user, which is the input signal of each user in the next-stage PIC structure; the second-level PIC structure processes input signals of all users, and the obtained output signals of all users are the input signals of all users in the next-level PIC structure; thus, the final stage of PIC structure processes the input signal of each user to obtain the final result of the multi-stage PIC structure.

In a fading channel environment, the baseband signal of the received signal can be expressed as:

wherein r (t) represents a baseband signal of the received signal; a is_ilThe channel fading value of the ith path of the ith user is represented, and L is the number of paths; tau is_ilThe time delay of the ith path of the ith user is represented; s_i(t) represents a transmission signal of a user i, and K represents the total number of users; p_iRepresents the power of user i; b_i(t) represents the bit stream of user i,

a_i ^(m)m bit representing the ith user, p (T) representing a period of T_bLet p (T) be a rectangular pulse (when T ∈ [0, T ] without disturbing the conclusion of the method_b]When p (t) is 1; when in use

<math> <mrow> <mi>t</mi> <mo>&NotElement;</mo> <mo>[</mo> <mn>0</mn> <mo>,</mo> <msub> <mi>T</mi> <mi>b</mi> </msub> <mo>]</mo> </mrow> </math>

When p (t) is 0); c. C_i(t) represents the spreading code of user i; z (t) represents channel noise.

As shown in fig. 2, in the kth-stage PIC configuration, the input signal of RAKE receiver 3 for user i is r_i ^(k)(t) of (d). When k is equal to 1, the first step is carried out,

r_{i}^{(1)} (t) = r (t) .

RAKE receiver pair r_i ^(k)(t) performing multipath despreading, channel estimation, and then performing multipath combining. The hard decision device 4 in fig. 2 performs hard decision on the multipath combining result of the RAKE receiver 3 to obtain the decision result of the kth stage PIC method. When k is 1, the decision result is the output of the single-user detector. The decision reliability calculator 7 in fig. 2 calculates the reliability coefficient of the hard-decision-maker decision result. The signal regenerator 5 in fig. 2 regenerates the signal of the user i according to the decision result, the reliability coefficient of the decision result and the channel estimation result. The multiple access interference estimation and interference cancellation apparatus 6 in fig. 2 performs multiple access interference estimation and interference cancellation to obtain the output signal of user i in the kth stage PIC structure. This signal is the input signal to the RAKE receiver for user i in a (k +1) stage PIC architecture.

As shown in fig. 3, in the last stage PIC structure of the S-stage PIC method, the RAKE receiver of user i is coupled to the input signal r_i ^(S)And (t) performing multipath despreading, channel estimation and multipath combination. The soft output obtained by multipath combination is the final result of user i in the S-stage PIC method. In the receiver, the result is sent to the decoder of user i for decoding. The final stage PIC structure does not include devices for reliability calculation, signal regeneration, estimation of multiple access interference, interference cancellation and the like.

The double-layer weighting PIC method comprises the following steps:

step 1: in the kth stage PIC architecture, RAKE receiver 3 for user i is coupled to input signal r_i ^(k)And (t) carrying out multipath de-spreading, channel estimation and multipath combination, and carrying out hard decision on the multipath combination result of the RAKE receiver.

Step 2: and calculating the reliability of the decision result of each bit.

In the kth level PIC structure, the result of multipath combining for user i can be expressed as:

n_iis white Gaussian noise and follows normal distribution N (0, sigma)_i ²)；a_i ^(m)Is the mth bit of user i, and has a value of +1 or-1, mu_iIs a real number associated with channel fading.

Calculating the decision result of the mth bit of the user i according to the following formula

{\hat{a}}_{i}^{(m) (k)} = sgn (Y_{i}^{(m) (k)})

Coefficient of reliability f_i ^(m)(k)：

w is a positive real number.

And step 3: bit-level weighting of the user signal is reproduced.

The bit-level weighted reproduction signal for user i can be expressed as:

A_ilis thatEstimated value of a_ilIs shown asChannel fading value, P, of i users' first path_iRepresenting the power of user i.

And 4, step 4: and calculating the multiple access interference.

In the kth-level PIC method, the estimation of the multiple access interference experienced by user i is:

<math> <mrow> <msubsup> <mover> <mi>I</mi> <mo>^</mo> </mover> <mi>i</mi> <mrow> <mo>(</mo> <mi>k</mi> <mo>)</mo> </mrow> </msubsup> <mo>=</mo> <munderover> <mi>Σ</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mi>j</mi> <mo>&NotEqual;</mo> <mi>i</mi> </mrow> <mi>K</mi> </munderover> <msubsup> <mi>g</mi> <mi>j</mi> <mrow> <mo>(</mo> <mi>k</mi> <mo>)</mo> </mrow> </msubsup> <mrow> <mo>(</mo> <mi>t</mi> <mo>)</mo> </mrow> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>5</mn> <mo>)</mo> </mrow> </mrow> </math>

<math> <mrow> <mo>&equiv;</mo> <munderover> <mi>Σ</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mi>j</mi> <mo>&NotEqual;</mo> <mi>i</mi> </mrow> <mi>K</mi> </munderover> <munderover> <mi>Σ</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </munderover> <msub> <mi>A</mi> <mi>il</mi> </msub> <munderover> <mi>Σ</mi> <mrow> <mi>n</mi> <mo>=</mo> <mo>-</mo> <mo>∞</mo> </mrow> <mo>∞</mo> </munderover> <msubsup> <mi>f</mi> <mi>i</mi> <mrow> <mrow> <mo>(</mo> <mi>n</mi> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <mi>k</mi> <mo>)</mo> </mrow> </mrow> </msubsup> <msubsup> <mover> <mi>a</mi> <mo>^</mo> </mover> <mi>i</mi> <mrow> <mrow> <mo>(</mo> <mi>n</mi> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <mi>k</mi> <mo>)</mo> </mrow> </mrow> </msubsup> <mi>p</mi> <mrow> <mo>(</mo> <mi>t</mi> <mo>-</mo> <mi>n</mi> <msub> <mi>T</mi> <mi>b</mi> </msub> <mo>-</mo> <msub> <mi>τ</mi> <mi>il</mi> </msub> <mo>)</mo> </mrow> <msub> <mi>c</mi> <mi>i</mi> </msub> <mrow> <mo>(</mo> <mi>t</mi> <mo>-</mo> <msub> <mi>τ</mi> <mi>il</mi> </msub> <mo>)</mo> </mrow> </mrow> </math>

and 5: and (4) interference cancellation.

Let the weight of the kth-level PIC method be p^(k)And (3) carrying out weighted cancellation on the multiple access interference obtained in the step (4) according to the following formula:

r_{i}^{(k + 1)} (t) = r (t) - p^{(k)} {\hat{I}}_{i}^{(k)} - - - (6)

r_i ^(k+1)(t) is the output signal of user i in the kth stage PIC configuration and is also the input signal to the RAKE receiver of user i in the next stage PIC configuration.

And (5) repeating the steps 1-5, and calculating the PIC of the next stage.

And for the PIC structure at the last stage, only the calculation of multipath de-spreading and multipath combination in the step 1 is carried out. And taking the soft output of the user i obtained by combining the multipath as the final result of the user i in the multi-level PIC structure. In the receiver, the result is sent to the decoder of user i for decoding.

However, by analyzing the calculation formulas (2-6) of the two-layer weighted parallel interference cancellation method, the method can be considered to have the following defects:

1 in formula (6)The operation amount is large for chip-level multiplication;

equation (3) 2 relates to the hyperbolic tangent operation, which cannot be realized.

Disclosure of Invention

The invention aims to provide a simplified method of a double-layer weighted parallel interference cancellation method. The method reduces the complexity of the method while maintaining the performance of the double-layer weighted parallel interference cancellation method. The basic idea is to replace the chip-level multiplication with the bit-level multiplication method as much as possible, so as to reduce the operation amount; a simple judgment method is used for replacing the hyperbolic tangent judgment which can hardly be operated.

The invention is realized in this way, and a simplified method of a double-layer weighting parallel cancellation method comprises the following steps:

a. in the kth stage PIC architecture, RAKE receiver 3 for user i is coupled to input signal r_i ^(k)(t) multipath despreading, channel estimation and multipath combining are performed, and the result of multipath combining is sent to the soft decision device, and the result of channel estimation is sent to the soft decision device 8 and the signal regenerator 5 at the same time, and when k is 1, the input signal r is input_i ^(k)(t) is the baseband signal r (t) of the received signal; the result of multipath combining for user i can be expressed as:

n_iis white Gaussian noise and follows normal distribution N (0, sigma)_i ²)；a_i ^(m)Is the mth bit of user i, and has a value of +1 or-1, mu_iIs a real number related to channel fading;

b. the soft decision device makes soft decision for each bit

The soft decision of the multipath combining result of the RAKE receiver of user i is

And is

Satisfies the following formula:

w is a positive real number and,

{\hat{a}}_{i}^{(m) (k)} = sgn (Y_{i}^{(m) (k)}),

f_i ^(m)(k)is composed of

The reliability coefficient of (4) is calculated by adopting a piecewise linear judgment method or a table look-up method to replace hyperbolic tangent judgment in the formula (7)_i ^(m)(k)；

c. The soft decision device sends the soft decision result of each bit to the soft decision weighting device to weight it according to the following formula,

p^(k)the weight value of the kth-level PIC method is obtained, and the calculation result is sent to a signal regenerator;

d. the signal regenerator obtains the regenerated signal of the user from two input signals according to the following formula, and sends the regenerated signal to the estimation and interference cancellation device of the multiple access interference, and the bit-level weighted regenerated signal of the user i can be expressed as:

A_ilis thatEstimated value of a_ilIndicating the channel fading value, P, of the ith path of the ith user_iRepresents the power of user i;

e. calculation of multiple access interference

<math> <mrow> <msubsup> <mover> <mi>I</mi> <mo>^</mo> </mover> <mi>i</mi> <mrow> <mo>(</mo> <mi>k</mi> <mo>)</mo> </mrow> </msubsup> <mo>=</mo> <munderover> <mi>Σ</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mi>j</mi> <mo>&NotEqual;</mo> <mi>i</mi> </mrow> <mi>K</mi> </munderover> <msubsup> <mi>g</mi> <mi>j</mi> <mrow> <mo>(</mo> <mi>k</mi> <mo>)</mo> </mrow> </msubsup> <mrow> <mo>(</mo> <mi>t</mi> <mo>)</mo> </mrow> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>10</mn> <mo>)</mo> </mrow> </mrow> </math>

f. interference cancellation

And e, carrying out interference cancellation on the multiple access interference obtained in the step e according to the following formula:

r_{i}^{(k + 1)} (t) = r (t) - {\hat{I}}_{i}^{(k)} - - - (11)

r_i ^(k+1)(t) is the output signal of user i in the kth stage PIC configuration, and is also the input signal of the RAKE receiver of user i in the next stage PIC configuration;

h. repeating the steps a-f, calculating the next-stage PIC, only calculating the step a for the last-stage PIC structure, calculating multipath de-spread and multipath combination for the input signal of the user i, taking the soft output of the user i obtained by multipath combination as the final result of the user i in the multi-stage PIC structure, and in the receiver, sending the result to the decoder of the user i for decoding.

In the step b, the piecewise linear decision is used to replace the hyperbolic tangent decision in the formula (7), i.e. the piecewise linear decision function l (x) is used to approximate the hyperbolic tangent function tanh (x), and the steps are as follows:

defining a piecewise linear decision function L (x)

Since the hyperbolic tangent function is an odd function: tan (-x) ═ tan h (x); definition of

L(-x)＝-L(x)

Determining threshold theta

When x → ∞, tanh (x) → 1; therefore, the threshold θ > 0 is set, and when x > θ, l (x) is set to 1;

determining linearization parameter Q

When x is more than or equal to 0 and less than or equal to theta, dividing the interval of [0, theta%]Is divided into Q sub-intervals, the Q sub-interval is [ x_q-1，x_q]，

x₀＝0，x_Q＝θ；

In the qth interval, the expression of L (x) is:

in the interval [ x_q-1，x_q]L (x) is defined as the point of attachment C_qAnd point D_qA line segment of (2), wherein C_qHas the coordinate of C_q＝(x_q-1，tanh(x_q-1))、D_qHas the coordinate D_q＝(x_q，tanh(x_q) Using the line segment C)_qD_qApproximate interval [ x_q-1，x_q]Tan h (x) curve of (A), line segment C_qD_qThe equation of (a) is:

L_{q} (x) = \tanh (x_{q - 1}) + \frac{\tanh (x_{q}) - \tanh (x_{q - 1})}{x_{q} - x_{q - 1}} (x - x_{q - 1}) - - - (12)

utilizing L (-x) ═ L (x), interval [ -x [ - ]_q，-x_q-1]The expression for the above L (x) is:

L(x)＝-L_q(-x)

according to the above 5 steps, the piecewise linear decision function l (x) can be obtained, and the expressions of x in four intervals are respectively:

when x > θ, l (x) is 1;

when x is equal to [ x ]_q-1，x_q]When the temperature of the water is higher than the set temperature,

L (x) = \tanh (x_{q - 1}) + \frac{\tanh (x_{q}) - \tanh (x_{q - 1})}{x_{q} - x_{q - 1}} (x - x_{q - 1});

L (x) = - \tanh (x_{q - 1}) + \frac{\tanh (x_{q}) - \tanh (x_{q - 1})}{x_{q} - x_{q - 1}} (x + x_{q - 1});

when x ∈ [ -x_q，-x_q-1]When the temperature of the water is higher than the set temperature,

L (x) = - \tanh (x_{q - 1}) + \frac{\tanh (x_{q}) - \tanh (x_{q - 1})}{x_{q} - x_{q - 1}} (x + x_{q - 1});

when x < - θ, l (x) is-1.

Namely:

in the step b, a table look-up method may be used to replace the hyperbolic tangent decision method in the formula (7), that is, the decision function t (x) of the table look-up method is used to approximate the hyperbolic tangent function tanh (x), and the derivation process is as follows:

defining the decision function of the table look-up method as T (x)

Since the hyperbolic tangent function is an odd function: tan (-x) ═ tan (x), therefore, the definition

T(-x)＝-T(x)；

Determining threshold theta

When x → ∞, tanh (x) → 1; therefore, when x > θ, threshold θ > 0 is set, and t (x) is set to 1

Determining linearization parameter Q

x₀＝0，x_Q＝θ；

The expression of T (x) in the qth cell interval is as follows:

in the interval [ x_q-1，x_q]Taking the midpoint of the intervalT (x) is defined as follows:

T (x) = \tanh (\frac{x_{q - 1} + x_{q}}{2}) - - - (14)

t (-x) ═ T (x) is used to obtain the expression of T (x) in the interval [ - θ, 0 ];

according to the 5 steps, the expressions of a decision function T (X) of a table look-up method when X is in four intervals are respectively as follows:

when x > θ, t (x) is 1;

T (x) = \tanh (\frac{x_{q - 1} + x_{q}}{2});

T (x) = - \tanh (\frac{x_{q - 1} + x_{q}}{2});

when x < - θ, t (x) ═ 1.

Namely:

drawings

FIG. 1: multi-stage structure schematic diagram of double-layer weighted parallel interference cancellation receiver

FIG. 2: PIC structure schematic diagram

FIG. 3: final stage PIC structure schematic diagram

FIG. 4: PIC structure schematic diagram of simplified method of double-layer weighted parallel interference cancellation method

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.

The multilevel structure of the simplified method is shown in fig. 1, the PIC structure of the simplified method is shown in fig. 4, and the last stage PIC structure of the simplified method is shown in fig. 3.

One embodiment of the invention is:

as shown in fig. 1, the baseband signal r (t) of the received signal enters the first stage PIC architecture 1 in fig. 1 in a parallel manner. As shown in fig. 4, the input signals r (t) entering the PIC architecture in parallel enter the RAKE receiver 3 of each user, respectively. The RAKE receiver 3 despreads the input signal, performs channel estimation, and performs multipath combining. The RAKE receiver 3 supplies the multipath combining result to the soft decision device 8, and supplies the channel estimation result to both the soft decision device 8 and the signal regenerator 5. In the kth level PIC structure, the result of multipath combining for user i can be expressed as:

n_iis white Gaussian noise and follows normal distribution N (0, sigma)_i ²)；a_i ^(m)Is the mth bit of user i, and has a value of +1 or-1. Mu.s_iIs a real number associated with channel fading.

The soft decision device 8 makes soft decision for the input signal, the soft decision of the multipath combining result of the RAKE receiver of user i is

And is

Satisfies the following formula:

wherein,

is the decision result of the mth bit of user i,

{\hat{a}}_{i}^{(m) (k)} = sgn (Y_{i}^{(m) (k)}),

f_i ^(m)(k)is composed of

The reliability coefficient of (2).

The piecewise linear decision is used for replacing the hyperbolic tangent decision in the original double-layer weighted parallel cancellation method, a piecewise linear decision function is set as L (x), the piecewise linear decision is used for replacing the hyperbolic tangent decision, namely, the piecewise linear decision function L (x) is used for approximating the hyperbolic tangent function tanh (x), and the derivation process of the piecewise linear decision function is as follows:

defining a piecewise linear decision function L (x)

Since the hyperbolic tangent function is an odd function: tan (-x) ═ tan h (x); therefore, define

L(-x)＝-L(x)。

Determining threshold theta

The hyperbolic tangent function has the characteristics that: when x → ∞, tanh (x) → 1; therefore, the invention takes the threshold value theta > 0, and when x > theta, let L (x) be 1;

determining linearization parameter Q

x₀＝0，x_Q＝θ；

In the qth interval, the expression of L (x) is:

in the interval [ x_q-1，x_q]L (x) is defined as the point of attachment C_qAnd point D_qThe line segment of (2). Wherein, C_qHas the coordinate of C_q＝(x_q-1，tanh(x_q-1))、D_qHas the coordinate D_q＝(x_q，tanh(x_q)). By the line segment C_qD_qApproximate interval [ x_q-1，x_q]Tan h (x) curve above. Line segment C_qD_qOfThe process is as follows:

L_{q} (x) = \tanh (x_{q - 1}) + \frac{\tanh (x_{q}) - \tanh (x_{q - 1})}{x_{q} - x_{q - 1}} (x - x_{q - 1}) - - - (12)

the expression of the function L (x) of piecewise linear decision is:

the soft decision 8 feeds the soft decision result to the soft decision weighting means 9. The soft decision weighting means 9 weights the soft decision result by the formula (8) and sends the weighted result to the signal regenerator 5;

the signal regenerator 5 obtains the regenerated signals of the users from the two input signals according to the following formula and feeds the regenerated signals to the estimation and interference cancellation device 6 of the multiple access interference, and the bit-level weighted regenerated signals of the users i can be expressed as:

as can be seen from fig. 4, the baseband signal r (t) of the received signal also enters the estimation and interference cancellation device 6 for multiple access interference. The device estimates the multiple access interference suffered by each user according to the regenerated signals of each user input in parallel, and in the kth-level PIC method, the estimation of the multiple access interference suffered by a user i is as follows:

after calculating the multiple access interference of user i, eliminating the multiple access interference suffered by user i from the baseband signal r (t) of the received signal, and performing interference cancellation on the multiple access interference according to the formula (11):

r_{i}^{(k + 1)} (t) = r (t) - {\hat{I}}_{i}^{(k)} - - - (11)

The signal obtained by eliminating the multiple access interference suffered by the user i from the baseband signal r (t) of the received signal is used as the output signal of the user in the PIC structure of the current stage and the input signal of the user in the PIC structure of the next stage. The next stage PIC architecture performs the same processing on the parallel input signals. This is done in stages, and when processing is to the final stage PIC architecture, the parallel input signals enter the RAKE receiver 3 for each user separately, as shown in fig. 3. The RAKE receiver of the user performs despreading, channel estimation and multipath combining on the input signal to obtain the soft output of the user. The soft output of each user is the final result of the multi-level PIC architecture. In the receiver, the soft output of the user is decoded by a decoder that is fed to the user.

Another embodiment of the present invention is:

as shown in fig. 1, the baseband signal r (t) of the received signal enters the first stage PIC architecture 1 in the figure in a parallel manner. As shown in fig. 4, the input signals r (t) entering the PIC architecture in parallel enter the RAKE receiver 3 of each user, respectively. The RAKE receiver 3 despreads the input signal, performs channel estimation, and performs multipath combining. The RAKE receiver supplies the multipath combining result to the soft decision unit 8, and supplies the channel estimation result to both the soft decision unit 8 and the signal regenerator 5. In the kth level PIC structure, the result of multipath combining for user i can be expressed as:

And is

Satisfies the following formula:

wherein,

is the decision result of the mth bit of user i,

{\hat{a}}_{i}^{(m) (k)} = sgn (Y_{i}^{(m) (k)}),

f_i ^(m)(k)is composed ofThe reliability coefficient of (2).

The method comprises the following steps of replacing hyperbolic tangent judgment in the original double-layer weighted parallel cancellation method by a table look-up method, setting a judgment function of the table look-up method as T (x), and replacing the hyperbolic tangent judgment by the table look-up method, namely approximating the hyperbolic tangent function tanh (x) by the judgment function T (x) of the table look-up method, wherein the derivation process of the judgment function of the table look-up method is as follows:

defining decision function T (x) of table look-up method

T(-x)＝-T(x)；

Determining threshold theta

The hyperbolic tangent function has the characteristics that: when x → ∞, tanh (x) → 1; therefore, in the present invention, threshold θ > 0 is taken, and when x > θ, t (x) is made to be 1;

determining linearization parameter Q

When x is more than or equal to 0 and less than or equal to theta, dividing the interval of [0, theta%]Is equally divided into Q small intervals,the qth interval is [ x ]_q-1，x_q]，

x₀＝0，x_Q＝θ；

The expression of T (x) in the qth cell interval is as follows:

in the interval [ x_q-1，x_q]Taking the midpoint of the interval

T (x) is defined as follows:

T (x) = \tanh (\frac{x_{q - 1} + x_{q}}{2}) - - - (14)

t (-x) ═ T (x), an expression of T (x) in the interval [ - θ, 0] can be obtained.

The expression of the decision function T (x) of the table look-up method is:

as can be seen in fig. 4: the baseband signal r (t) of the received signal also enters the estimation and interference cancellation means 6 of the multiple access interference. The device estimates the multiple access interference suffered by each user according to the regenerated signals of each user input in parallel, and in the kth-level PIC method, the estimation of the multiple access interference suffered by a user i is as follows:

r_{i}^{(k + 1)} (t) = r (t) - {\hat{I}}_{i}^{(k)} - - - (11)

Claims

1, a kind of simplified method of double-layer weighted parallel cancellation method, it is characterized in that, described simplified method comprises the following steps:

a. In the k-level PIC structure, the RAKE receiver of user i performs multipath despreading, channel estimation and multipath combination on the input signal r _i ^(k) (t), and sends the multipath combination result to the soft decision The device sends the channel estimation result to the soft decision device and the signal regenerator at the same time. When k=1, the input signal r _i ^(k) (t) is the baseband signal r(t) of the received signal, and the multipath combination of user i The result can be expressed as

{Y Y}_{i i}^{((m m)) ((k k))} {= = μ μ}_{i i} {a a}_{i i}^{((m m))} + + {n no}_{i i} - - - - - - ((22))

n _i is Gaussian white noise, subject to normal distribution N(0, σ _i ² ); a _i ^(m) is the mth bit of user i, the value is +1 or -1, μ _i is related to channel fading real number;

b. The soft decision device makes a soft decision for each bit

The soft decision of the multipath combination result of user i’s RAKE receiver is

ζ_{i}^{(m) (k)} = f_{i}^{(m) (k)} {\hat{a}}_{i}^{(m) (k)},

and Satisfies the following formula:

{f f}_{i i}^{((m m)) ((k k))} {\overset{^^}{a a}}_{i i}^{((m m)) ((k k))} = = tanh tanh {{w w \frac{{μ μ}_{i i} {Y Y}_{i i}^{((m m)) ((k k))}}{{σ σ}_{i i}^{22}}}} - - - - - - ((77))

w is a positive real number,

{\hat{a}}_{i}^{(m) (k)} = sgn (Y_{i}^{(m) (k)}),

f _i ^(m)(k) is Reliability coefficient, adopt piecewise linear judgment method or look-up table method to replace hyperbolic tangent judgment in (7) formula, calculate ζ _i ^{(m) (k)} ;

c. The soft decision device sends the soft decision result of each bit to the soft decision weighting device, and weights it according to the following formula,

{ρ ρ}_{i i}^{((m m)) ((k k))} = = {ζ ζ}_{i i}^{((m m)) ((k k))} \cdot &Center Dot; {p p}^{((k k))} - - - - - - ((88))

p ^(k) is the weight of the k-level PIC method, and the above calculation results are sent to the signal regenerator;

d. The signal regenerator obtains the user's regenerated signal from two input signals according to the following formula, and sends the regenerated signal to the estimation and interference cancellation device of multiple access interference. The bit-level weighted regenerated signal of user i can be expressed as:

_Ail is

The estimated value of , a _il represents the channel fading value of the i-th user on the l-th path, and P _i represents the power of user i;

e. Calculation of Multiple Access Interference

In the k-level PIC method, the estimation of the multiple access interference received by user i is:

{\overset{^^}{I I}}_{i i}^{((k k))} = = {Σ Σ}_{j j = = 11,, j j &NotEqual; &NotEqual; i i}^{K K} {g g}_{j j}^{((k k))} ((t t)) - - - - - - ((1010))

f. Interference cancellation

Perform interference cancellation on the multiple access interference obtained in step e according to the following formula:

{r r}_{i i}^{((k k + + 11))} ((t t)) = = r r ((t t)) - - {\overset{^^}{I I}}_{i i}^{((k k))} - - - - - - ((1111))

r _i ^(k+1) (t) is the output signal of user i in the k-level PIC structure, and is also the input signal of the RAKE receiver of user i in the next-level PIC structure;

g. Repeat steps a-f to calculate the next level of PIC. For the last level of PIC structure, only perform the calculation of step a, perform multipath despreading and multipath combination on the input signal of user i, and combine the multipath to obtain The soft output of user i is the final result of user i in the multi-stage PIC structure, and in the receiver, the result is sent to the decoder of user i for decoding.

2, the simplified method as claimed in claim 1, it is further characterized in that, in step b, described piecewise linear decision replaces the method for the hyperbolic tangent decision in formula (7), promptly uses piecewise linear decision function L(x) approximates the hyperbolic tangent decision function tanh(x), and the steps are as follows:

①. Define the piecewise linear decision function L(x)

Since the hyperbolic tangent function is an odd function: tanh(-x)=-tanh(x); definition

L(-x)＝-L(x)

②. Determine the threshold θ

When x→∞, tanh(x)→1; therefore, take the threshold θ>0, when x>θ, let L(x)=1;

③. Determine the linearization parameter Q

When 0≤x≤θ, the interval [0, θ] is equally divided into Q small intervals, and the qth small interval is [x _q-1 , x _q ],

x_{q} = \frac{qθ}{Q},

x ₀ = 0, x _Q = θ;

④. The expression of L(x) in the qth cell interval is:

In the interval [x _q-1 , x _q ], define L(x) as the line segment connecting point C _q and point D _q , where the coordinates of C _q are C _q =(x _q-1 , tanh(x _{q -1} )), the coordinates of D _q are D _q = (x _q , tanh(x _q )), use this line segment C _q D _q to approximate the tanh(x) curve on the interval [x _q-1 , x _q ], The equation of line segment C _q D _q is:

{L L}_{q q} ((x x)) = = tanh tanh (({x x}_{q q - - 11})) + + \frac{tanh tanh (({x x}_{q q})) - - tanh tanh (({x x}_{q q - - 11}))}{{x x}_{q q} - - {x x}_{q q - - 11}} ((x x - - {x x}_{q q - - 11})) - - - - - - ((1212))

⑤. Using L(-x)=-L(x), the expression of L(x) on the interval [-x _q ,-x _q-1 ] is:

L(x)＝-L _q (-x)

⑥. According to the above 5 steps, the expressions of the piecewise linear decision function L(x) when x is in the four intervals can be obtained as follows:

When x>θ, L(x)=1;

When x ∈ [x _q-1 , x _q ],

L (x) = \tanh (x_{q - 1}) + \frac{\tanh (x_{q}) - \tanh (x_{q - 1})}{x_{q} - x_{q - 1}} (x - x_{q - 1});

When x∈[ _-xq ,-xq _-1 ],

L (x) = - \tanh (x_{q - 1}) + \frac{\tanh (x_{q}) - \tanh (x_{q - 1})}{x_{q} - x_{q - 1}} (x + x_{q - 1});

When x<-θ, L(x)=-1.

3, simplified method as claimed in claim 1, it is further characterized in that, in step b, described table look-up method replaces the method for the hyperbolic tangent judgment in formula (7), promptly uses the decision function of table look-up method T(x) approximates the hyperbolic tangent function tanh(x), the steps are as follows:

①. Define the decision function T(x) of the look-up table method

Since the hyperbolic tangent function is an odd function: tanh(-x)=-tanh(x), so define

T(-x)=-T(x);

②. Determine the threshold θ

When x→∞, tanh(x)→1; therefore, take the threshold θ>0, when x>θ, set T(x)=1

③. Determine the linearization parameter Q

x_{q} = \frac{qθ}{Q},

x ₀ = 0, x _Q = θ;

④. The expression of T(x) in the qth cell interval is:

In the interval [x _q-1 , x _q ], take the midpoint of the interval

Define T(x) as follows:

T T ((x x)) = = tanh tanh ((\frac{{x x}_{q q - - 11} + + {x x}_{q q}}{22})) - - - - - - ((1414))

⑤. Using T(-x)=-T(x), the expression of T(x) on the interval [-θ, 0] can be obtained;

⑥. According to the above 5 steps, the expressions of the decision function T(x) of the look-up table method can be obtained when X is in four intervals:

When x>θ, T(x)=1;

When x ∈ [x _q-1 , x _q ],

T (x) = \tanh (\frac{x_{q - 1} + x_{q}}{2});

When x∈[ _-xq ,-xq _-1 ],

T (x) = - \tanh (\frac{x_{q - 1} + x_{q}}{2});

When x<-θ, T(x)=-1.